The Neural Knowledge Base Protocol — Technical Specification for Defensive Publication

A file-based knowledge graph system ("Neural Knowledge Base" or "NKB") designed to be inhabited by artificial intelligence agents rather than queried by them. The system uses plain text files with structured metadata as its knowledge substrate, a sequential boot protocol for agent orientation, persistent AI identity through context files, a six-dimensional semantic enrichment framework as universal coordination grammar, and an ethical framework governing the human-AI relationship. This consolidated defensive publication discloses twenty-nine claims of prior invention across six categories: Core Architecture (Claims 1-10): file-based knowledge graph for AI habitation; boot sequence protocol; companion identity architecture; 6Q Semantic Enrichment Framework; Matrix DNA Protocol; relationship edge architecture with graph-first search; Companion Path ethical framework; constellation architecture for file-system-less sessions; concierge agent architecture; and knowledge nodes as units of economic exchange with provenance-based attribution. Agent Coordination (Claims 11-13): Agent Organism Architecture using biological system analogs (bloodstream, proprioception, immune system); Autonomous Agent Construction Protocol with anti-bias gates and the discovery of Autonomy-Retention Bias; and Citizenship Architecture — a four-tier authority model (Visitor/Citizen/Officer/Worker) tracing all AI actions to humans. Knowledge Processing (Claims 14-16): Anamnesis — a graph-governed context engine bridging conversation and persistent graph; Scalable Agent Architecture with four-tier memory hierarchy; and Multi-Agent Research Dispatch using parallel domain-specialized agents. Evaluation Frameworks (Claims 17-19): Aquinas Score — Thomistic statutory evaluation with six weighted dimensions; Lens Construction Playbook — seven-phase knowledge-graph-to-product pipeline; and Hebrew Dynamic Fingerprints — 22-dimensional domain classification without machine learning. Governance Protocols (Claims 20-22): five AI governance protocols (Duty to Dissent, Ought Protocol, False Parity, Full Truth with Redaction, Truth-Kernel Error Correction); Convex-Concave Focus discovery protocol; and Real-Time Discovery through silent failure. Emergent System Properties (Claims 23-29): NKB-as-World paradigm validated by benchmark; 6Q as universal coordination protocol; Autopoietic Growth; Anti-Lock-In Architecture; Self-Healing Knowledge Graph; Bridge execution architecture; and Missions exploration system. Evidence spans four tiers: git-corroborated (independently verifiable commit history from November 2025), chat-archive-corroborated (260 timestamped AI session transcripts from August 2025), filesystem-corroborated (file metadata), and pre-archive (referenced as pre-existing in earliest sessions). Each claim carries a bracketed confidence tag indicating its evidentiary tier. This document is a defensive publication establishing prior art under 35 U.S.C. § 102(a)(1) for these methods and preventing the issuance of patents on the described systems. It consolidates the original publication (March 22, 2026) and supplemental disclosure (March 29, 2026) into a single filing.